opensidian was built predominantly with AI models. Nearly all of the code, the design docs, and this README were written by Claude (Anthropic’s models, via Claude Code) working from prompts, reviews, and corrections by a single human maintainer. The human decided what to build and what the trust model must guarantee, read and pushed back on the output, and ran it; the models wrote most of the lines. Commit messages record the design rationale in the same way — many were drafted by the model and edited by the maintainer.
yeah kudos! I personally think coming up with an MVP with LLMs are good. But then a human should continue the development till the first stable version.
So from what I have understood, it is entirely AI generated? Or was the prototype AI generated and then a human continued the way?
This is at the end of the README:
Holy balls, someone who actually disclosed the AI writing the program for them instead of lying and/or trying to hide it! Amazing!
yeah kudos! I personally think coming up with an MVP with LLMs are good. But then a human should continue the development till the first stable version.
Sounds like it was reverse engineered by an LLM
I think more like cloning the closed source software.